投资者交易的原因是什么外文翻译Word格式.docx

投资者交易的原因是什么外文翻译Word格式.docx

《投资者交易的原因是什么外文翻译Word格式.docx》由会员分享，可在线阅读，更多相关《投资者交易的原因是什么外文翻译Word格式.docx（7页珍藏版）》请在冰点文库上搜索。

原文

WhatMakesInvestorsTrade?

MaterialSource:

TheJournalOfFinance.Vol.Lvi.No.2.April2001

Author:

MarkGrinblattandMattiKeloharju

Abstract

Auniquedatasetallowsustomonitorthebuysells,andholdsofindividualsandinstitutionsintheFinnishstockmarketonadailybasis.Withthisdataset,weemployLogitregressionstoidentifythedeterminantsofbuyingandsellingactivityoveratwo-yearperiod.Wefindevidencethatinvestorsarereluctanttorealizelosses,theyengageintax-losssellingactivity,andthatpastreturnsandhistoricalpricepatterns,suchasbeingatamonthlyhighorlow,affecttrading.Therealsoismodest　evidencethatlife-cycletradingplaysaroleinthepatternofbuysandsells.

I.AUniqueDataSet

Thisstudyemploysacomprehensivedatasource:

thecentralregisterofshareholdingsforFinnishstocksintheFinnishCentralSecuritiesDepositoryFCSD.MostofthedetailsofthisdatasetarereportedinGrinblattandKeloharju（2000）.Forourpurposes,itisessentialtounderstandthat:

A.Thedataaggregateholdingsacrossbrokerageaccountsforthesameinvestor,whetherthesharesareheldinstreetnameornot.

B.Investorattributes,insubstantialdetail,arereportedwitheachtransaction.Amongthemoreinterestingattributesistheinvestorcategory.Weprimarilyfocusonfivecategories,basedonaclassificationsystemthathasbeendeterminedbytheEuropeanUnion,observedatthetopofTableI.AsixthforeigninvestorcategoryisaddedtotheanalysisofbuysversussellsinSectionIII.

C.Whenmultiplestockpurchasesoccur,wecomputethebasisfortheholding’scapitalgainorlossasthesharevolumeweighted-averagebasisproperlyadjustedforsplitsoftheinvestor’sinventoryofstockacquiredinthesampleperiod.Thus,aninvestorwhopurchases100sharesofNokiaAat600FIMonJanuary6,1995,andthen200sharesofNokiaAat900FIMonFebruary10,1995,wouldintheabsenceoffurtherpurchaseshaveabasisof800FIMinNokiaAafterFebruary10,1995.Asaleof150sharesofNokiaAonFebruary11,1995,bythissameinvestoristhusassumedtoconsistof50sharespurchasedpreviouslyonJanuary6and100sharespurchasedonFebruary10.AnyexistingholdingsofNokiaAonDecember27,1994,plusholdingsacquiredsinceDecember27,1994,forwhichnopurchasepriceisavailableneedtohavebeensoldbeforeFebruary11,1995,toestablishthisbasiscorrectly.WewouldexcludetheFebruary11salefromouranalysisifthiswerenotthecase.Thedatasetisobviouslylarge.Thereareapproximatelyonemillionselltransactionsandonemillionbuytransactionsthatweinitiallyscreen.Inaddition,forcomparisonpurposes,andconsistentwithOdean（1998）,mostofouranalysismatcheseachsellwithallstocksintheinvestor’sportfoliothatarenotsoldthesameday.Thus,ouranalysisofstocksalesbeginswithmillionsofevents.Severalfactors,outlinedbelowandspecifictothetypeofregressionundertaken,reducethesizeofthesampletothatreportedinourregressions.

D.InSectionII,whichreportstheresultsofregressionsthatstudysellversusholdbehavior,wenetallsamedaytradesinthesamestockbythesameinvestortomitigatetheeffectofintradaymarketmakinganddoublecountingduetotradesplitting,andwerequirethatthepurchasepriceusedtocomputethecapitalgainorlossforasaleorpotentialsalebeunambiguous.InSectionIII,whichstudiesbuyversussellbehavior,wenetintradaybuysandintradaysellsseparately,exceptfornominee-registeredforeigninvestorsforwhichthelackofpaneldatamakesnettingcomputationsimpossible.Finally,thereistherequirementthatallindependentvariablesbeavailableforallobservationswithinaninvestorcategory,butthishaslittleeffectonthesamplesize.

II.TheSellVersusHoldDecision

Thissectionanalyzesthedeterminantsofadummyvariablerepresentingthebinaryoutcome:

sellcodedasa“1”ordonotsellcodedasa“0”.Eachdaythataninvestorsellsstock,weexaminealloftheotherstocksinhisportfolioandclassifythemintooneofthesetwooutcomes,basedonwhetheranyofhisholdingsofthatstockweresold.WereportcoefficientsandtstatisticsfromaLogitregressionestimatedwithmaximumlikelihoodprocedures.WehaveverifiedthattheresultswewillreportshortlyareneitherNokiaspecificnoraffectedbyserialcorrelation,andthattheyaresimilartothoseobtainedfromthelesssensibleOLSspecification.

A.Descriptionoftheregression

Eachof293,034binarydatapoints,obtainedinthemannerdiscussedabove,belongstoaninvestorinoneofthefivedomesticinvestorclasses.Foreachdomesticinvestorclass,weestimatetherelationbetweenthedependentvariablesellversusholdand244regressors,ofwhich18areuniquetohouseholds,2areuniquetothefinanceandinsuranceinstitutions,and1isuniquetothegovernmentsector.Theseregressorsincludeasetofvariablesusedascontrolsforwhichcoefficientsarenotreported3andasetofreportedvariables.Thelatterinclude122variablesrelatedtopastreturns-listedinTableI,PanelA,whichanalyzespositivepastreturnsover11horizons,andPanelB,whichanalyzesnegativepastreturnsover11horizons,twodummyvariablesrepresentingmoderateandextremecapitallossesPanelC,twodummyvariablesrepresentingtheinteractionofaDecemberdummyandthecapitallossdummiesPanelD4theinteractionofadummyforaholding-periodcapitallossandthe22pastreturnvariablesPanelsEandF,tworeferencepricevariablestoassessifthesalesdecisionisaffectedbythestock’spricebeingataone-monthhighorlowPanelG,apairofvariablesrelatedtostockpriceandstockmarketvolatility~PanelH,andasetofeightmiscellaneousvariablesthatcontrolfortheinvestorandhisportfolioPanelI.Variablesrelatedtothesesetsofvariableshaveeitherbeenpostulatedtoberelatedtotrading,commonsensesuggeststheyshouldberelatedtotrading,ortheyhavebeenfoundinpriorempiricalresearchtoberelatedtotrading.Forexample,becausewecontrolforthestocktradedwithstockdummies,thestockvolatilityvariableaskswhetheraninvestortendstosellastockatatimewhenitsvolatilityishigherthannormal.

B.Pastreturnsandthesellversursholddecision

Theresultsarefairlyconsistentacrosstheinvestorcategories.Returnsbeyondamonthinthepast20tradingdaysappeartohavelittleimpactonthedecisiontosellversushold,whereaspositivemarket-adjustedreturnsonanydayofthelastweek,orduringthelastmonth,aresignificantlycorrelatedwiththedecisiontosell.Generally,themorerecentthepositivereturn,themorelikelyistheselldecision.Althoughtheresultsforday0arethestrongestofall,wedonothaveintradaypaneldatathatwouldallowustoseparateouttheimpactofreturnsontradingactivityfromtheimpactoftradingactivityonreturns.However,ifthereisasimultaneousequationsbias,itworkstobiasthecoefficientdownwards,andbecauseoftheday0returnwithalmostalloftheotherregressors,haslittleeffectontheothercoefficients.Weknowthisfromrunningouranalysiswithouttheday0regressors.PanelBindicatesthatinthepriorweek,themorenegativearethemarket-adjustedreturn,theloweristhepropensitytosell.ThesignificanceofthepositiveT-statisticsforhouseholdsandnon-financialcorporationsforhorizonsgoingbackuptooneweekpriortothesaleappearstobeweakerthantheimpactofthepositivereturnsonthepropensitytosell.Moreover,thereareoccasionalsignreversalsatsomeofthelongerhorizonsforsomeofthecategories.

ThisevidencesuggeststhatforFinnishinvestors,recentlargepositivemarket-adjustedreturnsuptoamonthinthepastareanimportantfactorintriggeringasell.Stronglynegativemarket-adjustedreturnsuptoaweekinthepasthaveamoderatetendencytoreducetheprobabilityofasell.

C.EvidenceontheDispositionEffect

Plottingthedistributionsofholdingperiodrealizedandpapercapitalgainsandlosseswithoutthecontrolsintheregressionisalsoinsightful.PanelAofFigure1showsthedistributionofrealizedgainsandlossesforallinvestorcategoriesaggregatedtogetherandPanelBshowsthepapergainsandlosses.ThelefttailofPanelA,therealizedcapitalgainreturns,ismuchthinnerthanthatinPanelB,thepapercapitalgainreturns.TherighttailinPanelAismuchthicker.PerhapsmoststrikingiswhatappearstobeadiscontinuityatzeroforPanelA’sdistributionofrealizedcapitalgainreturns.TotheleftofzeroinPanelA,theheightofthedensityfunctionimmediatelydropsoff.ForthepapercapitalgainreturnsofPanelB,thedistributiontotheleftofzeroappearstorelativelysmooth.

Althoughtheseplotslackhundredsofcontrolsfoundintheregressions,theyareconsistentwiththetendencyforlargegainstoberealizedandlargelossestobeheldonto.Theyalsotellastorythatisveryhardtoexplainasanythingbutadispositioneffect.Forexample,intheHarrisandRaviv（1993）model,investorshavebeliefsaboutacompany’sfutureprospectsthatarenotcloselytiedtostockprices.Hence,asstockpricedecline,stockinthatcompanybecomesmoreattractiveandviceversa.However,HarrisandRaviv’s（1993）modelisnotconsistentwiththediscontinuityobservedinFigure1,PanelA,butrather,withaskewedyetsmootherdistributionthanthatobserved.6

D.EvidenceonMiscellaneousStockandInvestorAttributesasDeterminantsofSales

Inadditiontopastreturns,capitallosses,tax-losssellingvariables,referencepriceeffects,andvolatility,ourregressionscontr